Selection of a suitable node to host a virtual machine in an environment containing a large number of nodes

• US 8,713,182 B2
• Filed: 08/03/2009
• Issued: 04/29/2014
• Est. Priority Date: 08/03/2009
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Agneeswaran, Vijay Srinivas
• Primary Examiner(s): Gilles, Jude Jean
• Assistant Examiner(s): Means, Jaren M
• Application Number:
  US12/534,157
• Publication Number:
  US 20110029672A1
• Time in Patent Office: 1,730 Days
• Field of Search: 709/226
• US Class Current: 709/226
• CPC Class Codes: G06F 9/5044 considering hardware capabilities
• CPC Class Codes: G06F 9/505 considering the load

Abstract

An aspect of the present invention facilitates selecting suitable nodes to host virtual machines (VMs) in an environment containing a large number of nodes (such as a grid). In one embodiment, information indicating corresponding resources available in each machine node (a node capable of hosting VMs) in the grid is maintained distributed over a set of management nodes contained in the grid. On receiving an indication that a VM requiring a set of resources is sought to be hosted, a machine node having available the set of resources is identified based on the distributed information. The VM is then provisioned/hosted on the identified machine node. The maintenance of the resource availability information distributed across multiple management nodes enables the solution to be scaled for use in environments having a large number of nodes.

Claims

1. A method of selecting suitable nodes to host virtual machines (VMs) in a grid containing a plurality of machine nodes capable of hosting VMs, said method comprising:

maintaining information indicating corresponding resources available for hosting VMs in each of said plurality of machine nodes,

wherein said information comprises resource availability data for each of said plurality of machine nodes and is maintained on a plurality of management nodes of said grid, said resource availability data for each machine node indicating the resources available in the machine node for hosting VMs,

wherein a specific management node of said plurality of management nodes for maintaining the resource availability data of a machine node is selected according to a convention which maps the resource availability data of the machine node to the specific management node whereby a respective portion of said information maintained by each management node comprises resource availability data for corresponding multiple machine nodes external to the management node;

receiving an indication that a first VM requiring a first set of resources is sought to be hosted on a suitable one of said plurality of machine nodes;

mapping, according to said convention, said first set of resources to a first management node of said plurality of management nodes;

identifying, by said first management node, a first machine node having available said first set of resources, wherein said first machine node is contained in said plurality of machine nodes, wherein said identifying comprises examining the corresponding portion of said information maintained in said first management node to identify said first machine node; and

provisioning said first VM on said first machine node as said suitable one of said plurality of machine nodes, after said identifying.

2. The method of claim 1, wherein said convention comprises a mathematical function that maps one or more values of resource availability data to corresponding identifiers of said plurality of management nodes,

wherein each management node of said plurality of management nodes maintains the resource availability data of machine nodes that is mapped to the identifier of the management node.

3. The method of claim 2, wherein said mathematical function comprises:

a first hash which generates a corresponding one of a set of hashed identifiers for each value of the available resources specified in said resource availability data; and

a second hash which generates a corresponding one of said set of hashed identifiers for each node identifier, wherein each node identifier uniquely identifies a corresponding one of said plurality of management nodes, wherein a first value of available resources specified in the resource availability data of a machine node is mapped to said first management node, if the same hashed identifier is generated by both of said first hash and said second hash based respectively on said first value and a node identifier of said first management node.

4. The method of claim 3,

wherein said indication is first received by said first management node of said plurality of management nodes,

wherein said identifying comprises:

checking in said first management node, by examining the respective portion maintained by said first management node, whether any one of the corresponding machine nodes has available all of said first set of resources;

if said portion maintained on said first management node indicates a machine node having available all of said first set of resources, providing the machine node as said first machine node;

if said portion maintained on said first management node indicates that there is no machine node having available all of said first set of resources,

forwarding said indication to a second management node; and

examining, in said second management node, the respective portion stored in said second management node to identify said first machine node as having available said first set of resources,

wherein said second management node is contained in said plurality of management nodes.

5. The method of claim 4, wherein said set of hashed identifiers are in a sequential order, wherein said convention further comprises:

if said first management node is unavailable, determining as said second management node a management node contained in said plurality of management nodes whose hashed identifier is a successor to the hashed identifier of said first management node according to said sequential order,

wherein said maintaining maintains said first value in said second management node if said first management node is unavailable, and

wherein said examining examines said second management node if said first management node is unavailable.

6. The method of claim 4, wherein said resource availability data for a machine node comprises a maximum size of a VM that can be supported by the machine node and a load factor indicating the load that can be sustained by the machine node under peak usage.

7. The method of claim 6, wherein said maximum size is specified as a combination of a measure of the processor time available in the machine node and an amount of memory available in the machine node.

8. The method of claim 7, wherein a value of available resource comprises a range of load factors and combinations of the measure of the processor time and the amount of memory.

9. The method of claim 8, wherein the value of available resource further comprises a geographical location of the machine node.

10. The method of claim 7, further comprising:

sending the resource availability data as a corresponding advertisement from each machine node to a corresponding management node,

wherein the corresponding management node determines the specific management node of said plurality of management nodes according to said convention, and causes said corresponding advertisement to be stored in the specific management node.

11. The method of claim 10, wherein the advertisement is in the form of a Resource Data Format (RDF) triple containing a subject, a predicate and an object,

wherein said maximum size is specified as the subject of said RDF triple, the load factor is specified as the predicate of said RDF triple and an identifier of said machine node is specified as the object of said RDF triple.

12. A computing system comprising:

a plurality of machine nodes capable of hosting virtual machines (VMs);

a plurality of management nodes operable to;

maintain information indicating corresponding resources available for hosting VMs in each of said plurality of machine nodes,

wherein said information comprises resource availability data for each of said plurality of machine nodes, said resource availability data for each machine node indicating the resources available in the machine node for hosting VMs,

wherein a specific management node for maintaining the resource availability data of a machine node is selected according to a convention which maps the resource availability data of the machine node to the specific management node whereby a respective portion for each management node comprises resource availability data for corresponding multiple machine nodes external to the management node;

receive an indication that a first VM requiring a first set of resources is sought to be hosted on a suitable one of said plurality of machine nodes;

map, according to said convention, said first set of resources to a first management node of said plurality of management nodes; and

identify, by said first management node, a first machine node having available said first set of resources, wherein said first machine node is contained in said set of machine nodes, wherein said identify comprises examining the corresponding portion of said information maintained in said first management node to identify said first machine node; and

a scheduler system to send said indication and to provision said first VM on said first machine node as said suitable one of said plurality of machine nodes, after said identify operation.

13. The computing system of claim 12, wherein said convention comprises a mathematical function that maps one or more values of resource availability data to corresponding identifiers of said plurality of management nodes,

wherein each management node of said plurality of management nodes maintains the resource availability data of machine nodes that is mapped to the identifier of the management node.

14. The computing system of claim 13, wherein said mathematical function comprises:

a first hash which generates a corresponding one of a set of hashed identifiers for each value of the available resources specified in said resource availability data; and

a second hash which generates a corresponding one of said set of hashed identifiers for each node identifier, wherein each node identifier uniquely identifies a corresponding one of said set of management nodes,

wherein a first value of available resources specified in the resource availability data of a machine node is mapped to said first management node, if the same hashed identifier is generated by both of said first hash and said second hash based respective on said first value and a node identifier of said first management node.

15. The computing system of claim 14,

wherein said indication is first received by said first management node of said set of management nodes,

wherein to perform said identify operation, said set plurality of management nodes are further operable to:

check in said first management node, by examining the respective portion maintained by said first management node, whether any one of the corresponding machine nodes has available all of said first set of resources;

if said portion maintained on said first management node indicates a machine node having available all of said first set of resources, provide the machine node as said first machine node;

if said portion maintained on said first management node indicates that there is no machine node having available all of said first set of resources,

forward said indication to a second management node; and

examine, in said second management node, the respective portion stored in said second management node to identify said first machine node as having available said first set of resources,

wherein said second management node is contained in said plurality of management nodes.

16. The computing system of claim 15, wherein said set of hashed identifiers are in a sequential order, wherein said convention further comprises:

if said first management node is unavailable, determining as said second management node a management node contained in said plurality of management nodes whose hashed identifier is a successor to the hashed identifier of said first management node according to said sequential order,

wherein said first value is maintained in said second management node if said first management node is unavailable, and

wherein said plurality of management nodes are operable to further examine said second management node if said first management node is unavailable.

17. The computing system of claim 15, wherein said resource availability data for a machine node comprises a maximum size of a VM that can be supported by the machine node and a load factor indicating the load that can be sustained by the machine node under peak usage,

wherein said maximum size is specified as a combination of a measure of the processor time available in the machine node and an amount of memory available in the machine node.

18. The computing system of claim 17, wherein the resource availability data is sent as a corresponding advertisement from each machine node to a corresponding management node,

wherein the corresponding management node determines the specific one of said set plurality of management nodes according to said convention, and causes said corresponding advertisement to be stored in the specific management node.

19. The computing system of claim 18, wherein the advertisement is in the form of a Resource Data Format (RDF) triple containing a subject, a predicate and an object,

wherein said maximum size is specified as the subject of said RDF triple, the load factor is specified as the predicate of said RDF triple and an identifier of said machine node is specified as the object of said RDF triple.

20. A non-transitory machine readable medium storing one or more sequences of instructions for causing a management node to facilitate selection of suitable nodes to host virtual machines (VMs) in a grid containing a plurality, of machine nodes capable of hosting VMs, wherein execution of said one or more sequences of instructions by one or more processors contained in said management node causes said management node to perform the actions of:

maintaining only a portion of total information indicating corresponding resources available for hosting VMs in each of said plurality of machine nodes,

wherein said total information comprises resource availability data for each of said plurality of machine nodes and is maintained on a plurality of management nodes of said grid, said plurality of management nodes including said management node, said resource availability data for each machine node indicating the resources available in the machine node for hosting VMs,

wherein a specific management node of said plurality of management nodes for maintaining the resource availability data of a machine node is selected according to a convention which maps the resource availability data of the machine node to the specific management node whereby a respective portion of said information maintained by each management node comprises resource availability data for corresponding multiple machine nodes external to the management node;

receiving a request to indicate a first machine node having available a first set of resources,

wherein said request is received in response to mapping, according to said convention, said first set of resources to said management node;

examining said portion to determine whether any one of the corresponding machines nodes has available all of said first set of resources;

if said examining determines a machine node as having available all of said first set of resources, sending a response indicating that said machine node is said first machine node having available said first set of resources;

if said examining determines no machine node as having available all of said first set of resources, forwarding said request to another management node of said plurality of management nodes.

21. The non-transitory machine readable medium of claim 20, wherein said convention comprises a mathematical function that maps one or more values of resource availability data to corresponding identifiers of said plurality of management nodes,

wherein each management node of said plurality of management nodes maintains the resource availability data of machine nodes that is mapped to the identifier of the management node.

22. The non-transitory machine readable medium of claim 21, wherein said mathematical function comprises:

a first hash which generates a corresponding one of a set of hashed identifiers for each value of the available resources specified in said resource availability data; and

a second hash which generates a corresponding one of said set of hashed identifiers for each node identifier, wherein each node identifier uniquely identifies a corresponding one of said plurality of management nodes,

wherein said portion comprises resource availability data for machine nodes having a first value of available resources, wherein the same hashed identifier is generated by both of said first hash and said second hash based respective on said first value and a node identifier of said management node,

wherein said another portion comprises resource availability data for machine nodes having a second value of available resources, wherein the same hashed identifier is generated by both of said first hash and said second hash based respective on said second value and a node identifier of said another management node.

23. The non-transitory machine readable medium of claim 22, wherein said set of hashed identifiers are in a sequential order, wherein said forwarding forwards said request to said another management node whose hashed identifier is a successor to the hashed identifier of said management node according to said sequential order.

24. The non-transitory machine readable medium of claim 23, wherein said resource availability data for a machine node comprises a maximum size of a VM that can be supported by the machine node and a load factor indicating the load that can be sustained by the machine node under peak usage, wherein said maximum size is specified as a combination of a measure of the processor time available in the machine node and an amount of memory available in the machine node.

25. The non-transitory machine readable medium of claim 24, wherein the resource availability data is received as a corresponding advertisement from each machine node in the form of a Resource Data Format (RDF) triple containing a subject, a predicate and an object,

wherein said maximum size is specified as the subject of said RDF triple, the load factor is specified as the predicate of said RDF triple and an identifier of said machine node is specified as the object of said RDF triple.

26. A method of selecting suitable nodes to host virtual machines (VMs) in a grid, said grid containing a plurality of machine nodes capable of hosting VMs, said method comprising:

receiving a plurality of advertisements, each advertisement being received from a machine node of said plurality of machine nodes, each advertisement containing resource availability data indicating the resources available in the machine node for hosting VMs;

for each resource availability data in said plurality of advertisements, selecting a corresponding management node according to a convention which maps one or more values constituting the resource availability data to one of a plurality of management nodes in said grid;

storing each of said plurality of advertisements in a corresponding selected management node such that the resource availability information of said plurality of machine nodes is maintained distributed in said plurality of management nodes;

receiving an indication that a first VM requiring a first set of resources is sought to be hosted on a suitable one of said plurality of machine nodes;

determining, according to said convention based on said first set of resources, a first management node of said plurality of management nodes as maintaining resource availability data of potential suitable machine nodes for hosting said first VM;

examining the corresponding portion of said resource availability information maintained in said first management node to identify a first machine node of said plurality of machine nodes as having available said first set of resources; and

provisioning said first VM on said first machine node as said suitable one of said plurality of machine nodes, after said determining and said examining.

27. The method of claim 26, wherein said convention comprises a mathematical function that maps one or more values of resource availability data to corresponding identifiers of said plurality of management nodes,

wherein each management node of said plurality of management nodes maintains the resource availability data of machine nodes that is mapped to the identifier of the management node.